Hello, was the GCGL results reported in the paper obtained with random_seed set to None? In this case, calling pipelin_xx separately during the training and testing phases will result in different training sets and testing sets, leading to incorrect evaluation results.
I think should set random_seed to the exact value and reevaluate the results of GCGL.
Best wishes
Hello, was the GCGL results reported in the paper obtained with random_seed set to None? In this case, calling pipelin_xx separately during the training and testing phases will result in different training sets and testing sets, leading to incorrect evaluation results.
I think should set random_seed to the exact value and reevaluate the results of GCGL.
Best wishes